Fourier domain mode locked (FDML) lasers are excellent tunable laser sources for frequency domain optical coherence tomography (FD-OCT) systems due to their combination of high sweep rates, large tuning ranges, and high output powers. However, conventional FDML lasers provide coherence lengths of only 4–10 mm, limiting their use in demanding applications such as intravascular OCT where coherence lengths of >20 mm are required for optimal imaging of large blood vessels. Furthermore, like most swept lasers, conventional FDML lasers produce only one useable sweep direction per tunable filter drive cycle, halving the effective sweep rate of the laser compared to the filter drive frequency. Here, we demonstrate a new class of FDML laser incorporating broadband dispersion compensation near 1310 nm. Elimination of chromatic dispersion in the FDML cavity results in the generation of forward (short to long wavelength) and backward (long to short wavelength) sweeps with substantially identical properties and coherence lengths of >21 mm. This advance enables long-range, high-speed FD-OCT imaging without the need for optical buffering stages, significantly reducing laser cost and complexity.
© 2011 OSA
Optical coherence tomography (OCT) is a depth-resolved imaging modality that provides micrometer-scale resolutions with millimeter-scale axial imaging ranges . Tunable laser sources are critical components of high-speed frequency domain OCT (FD-OCT) systems, and are the primary drivers of overall imaging system performance [2–4]. The sweep rate, tuning range, and coherence length of the laser determine the imaging speed, axial resolution, and imaging range of the FD-OCT system, respectively. The output power and noise of the laser source also strongly influence the sensitivity of the FD-OCT system. Fourier domain mode locked (FMDL) lasers  are commonly used in FD-OCT systems for academic research due to their ability to provide extremely high sweep rates of up to several MHz [6,7], broad tuning ranges of up to 200 nm [8,9], and output powers of up to 40 mW [6,7,10,11]. FDML lasers have shown promise for OCT applications in endoscopy , developmental biology , ophthalmology [6,13,14], and phase-sensitive photothermal imaging [15,16].
Although FDML lasers have been applied extensively in situations requiring moderate imaging ranges, they have not been widely deployed for long-range applications such as intravascular FD-OCT. Blood vessels are imaged using FD-OCT by introducing a rotary fiberoptic catheter [17–19], while a rapid injection of angiographic contrast fluid is used to temporarily displace the blood and provide a clear imaging field. Some human arteries have diameters of up to 5 mm. Accounting for eccentric catheter placement in the lumen, a desired penetration depth into the vessel wall of 2 mm, and using a refractive index of 1.4 for the contrast fluid and tissue, an FD-OCT imaging range of ~10 mm in air is required for optimal performance in this application. The tunable laser source should therefore ideally provide a coherence length of ~20 mm to ensure high system sensitivity over the entire imaging range. At wavelengths near 1550 nm, FDML lasers with >20mm coherence length have been demonstrated by applying a specifically designed sequence of standard single mode fiber (SMF) and dispersion compensation fiber (DCF) . For wavelength regions below 1300 nm, where the material dispersion of silica is positive, it is not possible to build low-loss DCFs. As a result, FDML lasers operating near 1310 nm have so far produced coherence lengths of only 4–10 mm [5,7,8], precluding widespread use in applications such as intravascular OCT.
Most wavelength swept lasers, including FDML lasers, incorporate a tunable filter element that is driven periodically over an alternating series of forward (short to long wavelength) and backward (long to short wavelength) scans. In FDML lasers, along with the majority of other FD-OCT tunable laser sources, the characteristics of the two sweep directions are quite different [4,21]. The forward sweep of a conventional FDML laser has a shorter coherence length and higher noise than the backward sweep, making it unusable for FD-OCT imaging . This reduces the maximum duty cycle of the laser to 50% and reduces the FD-OCT imaging speed by a factor of 2 compared to the sweep frequency of the tunable filter. The inability to exploit both sweep directions has necessitated the use of external optical buffering stages to replicate the useable sweep and recover the duty cycle wasted by the unusable sweep . Buffering stages require additional optical delay lines, polarization controllers, and booster semiconductor optical amplifiers (SOAs) that increase the size and cost of the laser system. Although these issues are not significant limitations for many research applications, they are detrimental when a portable, low-cost system is required for clinical use.
Here we report the development of a new class of FDML laser incorporating broadband dispersion compensation at a center wavelength of 1315 nm. Removal of chromatic dispersion in the FDML fiber delay line allows true synchronization between the filter sweep and the time-of-flight of light in the cavity over the entire tuning range of the laser. With dispersion compensation, an extended coherence length of >21 mm is obtained for both sweep directions, enabling the use of FDML lasers for long-range FD-OCT. Additionally, both the forward and backward sweeps generated by the laser produce equivalent noise, axial image resolution, and power, allowing both sweep directions to be used for FD-OCT imaging and precluding the need for external buffering stages.
2. Group delay dispersion in FDML cavities
FDML lasers overcome the fundamental limitations in sweep speed common to other types of tunable lasers by synchronizing the tuning rate of the intracavity filter element to the round-trip time of light in the cavity . When the FDML synchronization condition is met, light at a given wavelength propagates once around the cavity in precisely the amount of time required for the tunable filter to move over an integer number of sweep periods. This synchronization condition is shown in Eq. (1), where is a non-zero integer, is the tuning period of the filter, and is the group round-trip time of light through the cavity.
In the absence of chromatic dispersion, and the tunable filter appears stationary to each wavelength in the sweep regardless of the tuning frequency. In the presence of chromatic dispersion, however, the synchronization condition is met only for a single wavelength when the round-trip time is matched at the zero dispersion point of the fiber. In the best case, the round-trip time can be matched at a wavelength away from the zero dispersion point by slightly changing the FDML drive frequency, giving true synchronization at exactly two wavelengths.
Dispersion in the single-mode optical fiber used to form FDML cavities is known to cause severe degradation in the tuning range and coherence length of lasers operating at 1060 nm [6,11,13]. Dispersion compensating fiber has been shown to dramatically improve FDML performance at 1550 nm . For FDML lasers operating near 1310 nm, however, it is not possible to design a DCF based on waveguide dispersion of the fiber core, which is the standard way to provide an inverse sign of GVD. Photonic crystal fibers (PCF) can be applied for this purpose , but these devices typically exhibit very high propagation losses. Furthermore, use of DCF or PCF is very challenging when compensation is required for several kilometers of fiber over spectral ranges of >100 nm, since the effects of higher-order dispersion increase as the spectral coverage is extended.
Figure 1A shows the calculated group delay relative to the zero dispersion wavelength for a 2.04 km length of SMF-28e + Nexcore optical fiber (Corning Inc) assuming a zero dispersion wavelength of 1317 nm and a zero dispersion slope of 0.088 ps/nm2km. Wavelengths at the blue edge of a ±70 nm tuning range undergo 550 ps of additional delay relative to the center wavelength of 1317 nm, while wavelengths at the red edge of the tuning range undergo 340 ps of additional delay. This group delay mismatch translates into additional insertion losses as the delayed wavelengths are transmitted through the filter.
To theoretically describe the effects of dispersion compensation, we apply a simple estimate of the possible number of round trips of light at a certain wavelength in the cavity. Biedermann et al.  showed that the lower limit for the number of round trips in an FDML cavity can be estimated as:
In Eq. (2), F is the tunable filter finesse, G is the effective gain of the cavity, α is the fractional tuning range of the filter compared to its free spectral range (FSR), f is the tuning frequency, and is the relative group delay mismatch at wavelength λ compared to the center wavelength of the sweep. Effective gain is the difference between the small-signal gain of the SOA and the total loss of all components in the cavity, neglecting wavelength dependence for the purposes of this simplified model. Figure 1B shows the estimated number of round trips for an FDML laser operating at a drive frequency of 100 kHz with a filter finesse of 1000, an effective gain of 12 dB, and a fractional tuning range of 0.95. Values for are as shown in Fig. 1A. Even though the laser is operating near the zero dispersion point of the fiber, the number of expected round trips through the filter decreases by roughly 5 orders of magnitude over a 100 nm tuning range. This phenomena limits the coherence length of the laser and also causes axial resolution degradation as FD-OCT imaging lengths are increased.
3. Chirped fiber Bragg grating dispersion compensation modules
To overcome the effects of chromatic dispersion in FDML laser cavities, a series of non-linearly chirped fiber Bragg grating (CFBG) pairs were designed and fabricated by TeraXion Inc. (Quebec City, Canada). Each pair contained one positively-chirped and one negatively-chirped grating. The combination of the CFBG pair compensates both normal and anomalous chromatic dispersion, on each side of the zero dispersion wavelength. The gratings were fusion-spliced to a 4-port optical circulator as shown in Fig. 2A to form a complete dispersion compensation module (DCM) and were packaged in a housing measuring 165 x 75 x 10 mm. The measured group delay of the individual CFBGs along with the total group delay for the combined grating pair is shown in Fig. 2B. The measured residual group delay mismatch between this DCM and a 2.04 km fiber spool is shown in Fig. 2C. The raw data is shown along with a curve fit to remove measurement noise. The dispersion measurement was carried out with a setup as described by Wieser et al. . The peak group delay mismatch was reduced from 300 ps to better than ±6 ps over a 100 nm measurement range. This residual mismatch could be further reduced in the future by better optimization of the CFBG design to match the dispersion function of the optical fiber. Mechanical or thermal wavelength tuning could also be employed to fine-tune the zero dispersion wavelength of the DCM for even greater compensation accuracy. The DCM had an insertion loss of 3.2 dB and a 3 dB bandwidth of 137 nm, which was limited by the physical length of the gratings and the amount of group delay at the edges of the compensation band.
With chromatic dispersion substantially suppressed, the number of cavity round trips for each wavelength in the sweep is expected to increase. Figure 1B shows the estimated number of round trips in the laser when the DCM is used according to Eq. (2). The calculation is given for the actual measured residual group delay, as well as for the curve fit to the measured data. When the DCM is included in the cavity, the number of round trips falls by 2 orders of magnitude less than without the DCM. Both the average number of round trips as well as the spectral uniformity of the number of round trips is improved. This increased spectral uniformity suggests that dispersion compensation should improve overall coherence length as well as axial resolution as a function of imaging range. It should be noted again that this theoretical model is highly simplified, and that in an actual FDML laser many more processes govern the linewidth and coherence properties [24,25].
4. FDML design and characterization
The extended coherence length FDML laser schematic is shown in Fig. 3 . A broadband SOA (Thorlabs) is connected to a 60/40 splitter that taps out 40% of the cavity power. The remaining 60% of the light enters the DCM, followed by a fiber Fabry-Perot tunable filter (FFP-TF, LambdaQuest) with a finesse of 1000. A polarization controller is used to align the polarization state of the light to the high-gain axis of the SOA. A circulator, fiber spool of length 1.02 km, and Faraday mirror form a sigma-ring cavity to complete the system. The laser was driven at a frequency of 100 kHz, generating alternating forward and backward sweeps at a rate of 200 kHz. The average output power was 35 mW. The 10 dB full-width tuning range was 115 nm, which was limited by the FSR of the FFP-TF. The duty cycle was 100% taking into account both the forward and backward sweep directions.
The coherence length of the laser was measured by evaluating the amplitude rolloff in FD-OCT point spread functions as the imaging range was increased. First, a 50/50 optical splitter was connected to the output of the laser. A Mach-Zehnder interferometer (MZI) with a variable path mismatch was connected to one of the splitter output arms. Interference fringes were acquired at OCT imaging range intervals of 0.5 mm (1.0 mm MZI mismatches) over a span of 0.25–13.5 mm. The fringes were detected using a 1 GHz unbalanced photodiode connected to a transimpedance amplifier, and were stored on an 8-bit digital oscilloscope with a 500 MHz analog bandwidth. Following acquisition, the MZI data was corrected for the nonuniform frequency response of the receiver as measured using a radiofrequency spectrum analyzer. A separate interference fringe from a fixed MZI was collected simultaneously with every data set in order to provide a reference for k-space recalibration. Recalibration was performed by calculating the optical frequency spacing of consecutive samples in the reference fringe, and then resampling the OCT fringe according to this spacing.
FD-OCT point spread functions generated by the dispersion-compensated FDML laser are shown in Fig. 4 for the forward (Fig. 4A) and backward (Fig. 4B) sweep directions. For comparison, an identical data set was acquired after replacing the DCM with an isolator. The point spread functions generated by this uncompensated FDML laser are shown in Fig. 5 for the forward (Fig. 5A) and backward (Fig. 5B) sweep directions. The noise floor of the dispersion-compensated laser is 3.1 dB lower than the noise floor of the conventional FDML laser.
Sensitivity rolloff plots were generated by fitting polynomial curves to the point spread function amplitudes. As shown in Fig. 6A , the sensitivity rolloff of the laser is substantially improved when the DCM is incorporated into the system. The 6 dB rolloff point for the forward and backward scans are reached at imaging ranges (all measurements in air) of 10.5 mm and 10.7 mm respectively, giving a coherence length of ~21.2 mm. This coherence length measurement may be affected by the analog bandwidth limit of the oscilloscope (500 MHz), as the interference fringe frequency at the deepest imaging range reached 460 MHz and a rolloff in noise floor can be observed beginning at about 12 mm. For the conventional FDML laser without the DCM, the 6 dB rolloff point for the forward and backward scans are reached at imaging ranges of 5.7 mm and 6.2 mm respectively, giving a coherence length of 11.4 mm and 12.4 mm. The difference in sensitivity for the two sweep directions of the uncompensated laser is seen to be larger than for the compensated laser, especially at imaging ranges above 6 mm.
Figure 6B shows the measured axial resolution (full width at half maximum) in air obtained as a function of imaging depth for both the dispersion compensated and uncompensated FDML lasers. Over the first 10 mm of imaging range, the laser incorporating the DCM provided average axial resolutions of 13.8 um and 16.0 um for the forward and backward sweeps, respectively. Over the same range, the conventional FDML laser provided average axial resolutions of 22.5 um and 21.0 um for the forward and backward sweeps, respectively. The resolution of the conventional laser, however, begins to deteriorate beyond imaging ranges of 4 mm and is almost completely degraded beyond 10 mm. This effect is even more pronounced for the forward scan direction. When the DCM is added to the laser, the resolution is nearly constant from 0 mm to 12 mm of imaging range. Since both the axial resolution and coherence length are equivalent between the two sweep directions, the forward and backward sweeps can both be used for OCT imaging.
5. Imaging performance
To evaluate the imaging performance of the dispersion-compensated FDML laser design, a second laser was constructed that operated at a drive frequency of 50 kHz and produced alternating forward and backward sweeps at a rate of 100 kHz. This modification was made by swapping in a 2.04 km fiber spool in place of the 1.02 km spool, and by replacing the DCM with a second version designed to compensate for the longer fiber length. This laser had very similar performance to the laser described above, except the tuning range was reduced to 110 nm due to a reduced dispersion compensation bandwidth imposed by the longer fiber length. A commercially-available C7XR intravascular FD-OCT system (LightLab Imaging, a St. Jude Medical subsidiary) was modified to interface with the laser. The line rate of the laser was reduced from 200 kHz to 100 kHz in order to keep the interference fringe frequencies within the Nyquist-limited band imposed by the commercial system’s digitizer. A probe interface unit (PIU) containing rotary and pullback mechanisms and a Dragonfly fiberoptic catheter (LightLab Imaging, a St. Jude Medical subsidiary) were used to obtain volumetric spiral-scanned data sets. The Dragonfly catheter has a working distance of approximately 2 mm, a focal spot size of approximately 25 um, and an outer diameter of 0.9 mm. The lens is formed by fusion splicing segments of coreless fiber to graded-index fiber.
Images of two human fingerpads immersed in water are shown in Fig. 7A and 7B. Images were acquired at a rotational speed of 100 Hz and a line rate of 100 kHz. The catheter is visible at the center of the image. Due to the extended coherence length of the source, image brightness remains high even when the fingers are pulled away from the catheter by 2–3 mm (2.8–4.2 mm air equivalent). Since no complex conjugate removal techniques are necessary to artificially double the imaging range, the images are free from ghost artifacts and crosstalk. A stent phantom was also volumetrically imaged by pulling back the catheter to conduct a spiral scan protocol. The phantom was filled with water and had a diameter of 4 mm (5.6 mm air equivalent). The catheter was positioned such that it was completely opposed against one side of the phantom at the beginning of the pullback and completely opposed against the other side at the end of the pullback. A longitudinal cut through the data set is shown in Fig. 7C. As the ranging depth between the catheter core and the phantom wall varied from 0 to 5.6 mm (air equivalent) the wall brightness remained consistent due to the extended coherence length of the laser. A 3D reconstruction of the stent phantom is shown in Fig. 7D. A virtual cut has been made in the image volume to permit visualization of the interior of the phantom. Stent struts and cross-linkages are visible as dark bands on the phantom wall.
The inclusion of broadband dispersion compensation elements for FDML lasers operating near 1310 nm results in significant performance improvements. First, the coherence length is extended by almost a factor of two, from 11.4 to 12.4 mm without use of a DCM to >21 mm with the use of a DCM. Second, the axial resolution of both sweep directions becomes nearly invariant over an imaging range of 0–10 mm. Third, the noise floor amplitude is reduced by >3 dB when chromatic dispersion is removed from the cavity. This combination of long coherence length, constant axial resolution, and reduced noise floor make dispersion-compensated FDML lasers well-suited for use in demanding applications such as intravascular imaging. The additional benefit of 100% duty cycle without the use of external buffering stages serves to reduce system size, cost and complexity.
W. Wieser and R. Huber would like to acknowledge support from Prof. W. Zinth at the LMU Munich. Their work was supported by the German Research Foundation (DFG) within the Emmy Noether program (HU 1006/2-1), by the European Union project FUN OCT (FP7 HEALTH, Contract No. 201880) and the ERC Starting Grant FDML-Raman (FP7 ERC, contract no. 259158).
References and links
1. D. Huang, E. A. Swanson, C. P. Lin, J. S. Schuman, W. G. Stinson, W. Chang, M. R. Hee, T. Flotte, K. Gregory, C. A. Puliafito, and J. G. Fujimoto, “Optical coherence tomography,” Science 254(5035), 1178–1181 (1991). [CrossRef] [PubMed]
2. B. Golubovic, B. E. Bouma, G. J. Tearney, and J. G. Fujimoto, “Optical frequency-domain reflectometry using rapid wavelength tuning of a Cr4+:forsterite laser,” Opt. Lett. 22(22), 1704–1706 (1997). [CrossRef] [PubMed]
4. R. Huber, M. Wojtkowski, K. Taira, J. G. Fujimoto, and K. Hsu, “Amplified, frequency swept lasers for frequency domain reflectometry and OCT imaging: design and scaling principles,” Opt. Express 13(9), 3513–3528 (2005). [CrossRef] [PubMed]
5. R. Huber, M. Wojtkowski, and J. G. Fujimoto, “Fourier Domain Mode Locking (FDML): A new laser operating regime and applications for optical coherence tomography,” Opt. Express 14(8), 3225–3237 (2006). [CrossRef] [PubMed]
6. T. Klein, W. Wieser, C. M. Eigenwillig, B. R. Biedermann, and R. Huber, “Megahertz OCT for ultrawide-field retinal imaging with a 1050 nm Fourier domain mode-locked laser,” Opt. Express 19(4), 3044–3062 (2011). [CrossRef] [PubMed]
7. W. Wieser, B. R. Biedermann, T. Klein, C. M. Eigenwillig, and R. Huber, “Multi-megahertz OCT: High quality 3D imaging at 20 million A-scans and 4.5 GVoxels per second,” Opt. Express 18(14), 14685–14704 (2010). [CrossRef] [PubMed]
8. D. C. Adler, Y. Chen, R. Huber, J. Schmitt, J. Connolly, and J. G. Fujimoto, “Three-dimensional endomicroscopy using optical coherence tomography,” Nat. Photonics 1(12), 709–716 (2007). [CrossRef]
9. J. Zhang, G. J. Liu, and Z. P. Chen, “Ultra broad band Fourier domain mode locked swept source based on dual SOAs and WDM couplers,” Proc. SPIE 7554, 75541I, 75541I-5 (2010). [CrossRef]
10. Y. X. Mao, C. Flueraru, S. D. Chang, and S. Sherif, “High-power 1300 nm FDML swept laser using polygon-based narrowband optical scanning filter,” Proc. SPIE 7168, 716822, 716822-8 (2009). [CrossRef]
11. S. Marschall, T. Klein, W. Wieser, B. R. Biedermann, K. Hsu, K. P. Hansen, B. Sumpf, K. H. Hasler, G. Erbert, O. B. Jensen, C. Pedersen, R. Huber, and P. E. Andersen, “Fourier domain mode-locked swept source at 1050 nm based on a tapered amplifier,” Opt. Express 18(15), 15820–15831 (2010). [CrossRef] [PubMed]
12. M. W. Jenkins, D. C. Adler, M. Gargesha, R. Huber, F. Rothenberg, J. Belding, M. Watanabe, D. L. Wilson, J. G. Fujimoto, and A. M. Rollins, “Ultrahigh-speed optical coherence tomography imaging and visualization of the embryonic avian heart using a buffered Fourier Domain Mode Locked laser,” Opt. Express 15(10), 6251–6267 (2007). [CrossRef] [PubMed]
13. R. Huber, D. C. Adler, V. J. Srinivasan, and J. G. Fujimoto, “Fourier domain mode locking at 1050 nm for ultra-high-speed optical coherence tomography of the human retina at 236,000 axial scans per second,” Opt. Lett. 32(14), 2049–2051 (2007). [CrossRef] [PubMed]
14. V. J. Srinivasan, D. C. Adler, Y. L. Chen, I. Gorczynska, R. Huber, J. S. Duker, J. S. Schuman, and J. G. Fujimoto, “Ultrahigh-speed optical coherence tomography for three-dimensional and en face imaging of the retina and optic nerve head,” Invest. Ophthalmol. Vis. Sci. 49(11), 5103–5110 (2008). [CrossRef] [PubMed]
15. D. C. Adler, R. Huber, and J. G. Fujimoto, “Phase-sensitive optical coherence tomography at up to 370,000 lines per second using buffered Fourier domain mode-locked lasers,” Opt. Lett. 32(6), 626–628 (2007). [CrossRef] [PubMed]
16. C. Zhou, T. H. Tsai, D. C. Adler, H. C. Lee, D. W. Cohen, A. Mondelblatt, Y. H. Wang, J. L. Connolly, and J. G. Fujimoto, “Photothermal optical coherence tomography in ex vivo human breast tissues using gold nanoshells,” Opt. Lett. 35(5), 700–702 (2010). [CrossRef] [PubMed]
17. G. J. Tearney, M. E. Brezinski, B. E. Bouma, S. A. Boppart, C. Pitris, J. F. Southern, and J. G. Fujimoto, “In vivo endoscopic optical biopsy with optical coherence tomography,” Science 276(5321), 2037–2039 (1997). [CrossRef] [PubMed]
19. H. Yabushita, B. E. Bouma, S. L. Houser, H. T. Aretz, I. K. Jang, K. H. Schlendorf, C. R. Kauffman, M. Shishkov, D. H. Kang, E. F. Halpern, and G. J. Tearney, “Characterization of human atherosclerosis by optical coherence tomography,” Circulation 106(13), 1640–1645 (2002). [CrossRef] [PubMed]
20. B. R. Biedermann, W. Wieser, C. M. Eigenwillig, T. Klein, and R. Huber, “Dispersion, coherence and noise of Fourier domain mode locked lasers,” Opt. Express 17(12), 9947–9961 (2009). [CrossRef] [PubMed]
21. A. Bilenca, S. H. Yun, G. J. Tearney, and B. E. Bouma, “Numerical study of wavelength-swept semiconductor ring lasers: the role of refractive-index nonlinearities in semiconductor optical amplifiers and implications for biomedical imaging applications,” Opt. Lett. 31(6), 760–762 (2006). [CrossRef] [PubMed]
22. R. Huber, D. C. Adler, and J. G. Fujimoto, “Buffered Fourier domain mode locking: Unidirectional swept laser sources for optical coherence tomography imaging at 370,000 lines/s,” Opt. Lett. 31(20), 2975–2977 (2006). [CrossRef] [PubMed]